|
|
Network server performance when communicating with a great number of clients
Mašín, Jan ; Pelka, Tomáš (referee) ; Komosný, Dan (advisor)
The aim is to create a simulation system between two computers of the server-client type to generace the process to server while usany UDP protocol: to set different parameters of the generated process (size of paquet, interval of sending) and evaluace the server-sources consumption (memory), to work out the outcomes in a well-organized manner. There have been tested two servers (each of different output) and operational systéme Windows XP, Windows Server 2003, Linux Mandriva 8.0 and Linux Fedora 7, each of these had been installed on the servers with different outputs. The generated process is carried out with usaje of the paquet generator made in the programming language C++.Two different programmes needed have been created, one of them for klient(paquet generation in OS Windows), its aim is to generate the paquets, the sekond on efor server (functional version working in OS Linux had had to be made) receiving the generated paquets. Evaulation of the source-consumption was carried out by two monitoring programs, one of them suited for OS Windows, the other for OS Linux. Both programs safe the information about procesor-work and operational memory into a text dokument which is very simple and fast way for searching the information. The measured values show charts and graphs that express the efficiency of separate OSs or, in case of need, servers of different outputs are able to handle the generated paquets and how do the alterations of the paquet generator affect the source efficiency( paquet size and paquet-sending interval).
|
|
|
Localization of nodes in Internet using Vivaldi system with adaptive time step
Mašín, Jan ; Balej, Jiří (referee) ; Komosný, Dan (advisor)
The aim of this thesis was to identification with the principles of logical evaluation of the position of stations on the Internet. Read up on the localization algorithm called Vivaldi with adaptive time step and subsequently to its implementation in the operating system GNU/Linux CentOS distribution. Do one's homework the PlanetLab experimental network (http://www.planet-lab.org/). At selected stations from the network transfer created by the application and verify its function on the real servers located at various places around the globe and assess the accuracy achieved by estimating the distance between stations on the PlanetLab network. In this scope of activity, the application was created to measure the delay prediction using Vivaldi algorithm with adaptive time step which is on principle of operation a client-server where the client performs the steps of the algorithm, Vivaldi and the server only listens, collects the resulting data Vivaldi algorithm and stores them neatly file. Furthermore, the application was developed for direct measurement of the delay, which also functions as a client-server. These applications have been transferred to the selected nodes from the PlanetLab experimental network. Subsequently, these nodes were running, to carry out the necessary measurements. The resulting values were work into tables of using Microsoft Excel. These values were then compared with direct measurements and competitive positioning by the King. Vivaldi localization methods with adaptive time step and the King, were compared based on calculated estimates of both real estate errors and measurement using distribution function of the relative errors of both methods. All this information was evaluated to compare accuracy of both the localization methods and direct measurements.
|
|
|
Network server performance when communicating with a great number of clients
Mašín, Jan ; Pelka, Tomáš (referee) ; Komosný, Dan (advisor)
The aim is to create a simulation system between two computers of the server-client type to generace the process to server while usany UDP protocol: to set different parameters of the generated process (size of paquet, interval of sending) and evaluace the server-sources consumption (memory), to work out the outcomes in a well-organized manner. There have been tested two servers (each of different output) and operational systéme Windows XP, Windows Server 2003, Linux Mandriva 8.0 and Linux Fedora 7, each of these had been installed on the servers with different outputs. The generated process is carried out with usaje of the paquet generator made in the programming language C++.Two different programmes needed have been created, one of them for klient(paquet generation in OS Windows), its aim is to generate the paquets, the sekond on efor server (functional version working in OS Linux had had to be made) receiving the generated paquets. Evaulation of the source-consumption was carried out by two monitoring programs, one of them suited for OS Windows, the other for OS Linux. Both programs safe the information about procesor-work and operational memory into a text dokument which is very simple and fast way for searching the information. The measured values show charts and graphs that express the efficiency of separate OSs or, in case of need, servers of different outputs are able to handle the generated paquets and how do the alterations of the paquet generator affect the source efficiency( paquet size and paquet-sending interval).
|
|
|
Localization of nodes in Internet using Vivaldi system with adaptive time step
Mašín, Jan ; Balej, Jiří (referee) ; Komosný, Dan (advisor)
The aim of this thesis was to identification with the principles of logical evaluation of the position of stations on the Internet. Read up on the localization algorithm called Vivaldi with adaptive time step and subsequently to its implementation in the operating system GNU/Linux CentOS distribution. Do one's homework the PlanetLab experimental network (http://www.planet-lab.org/). At selected stations from the network transfer created by the application and verify its function on the real servers located at various places around the globe and assess the accuracy achieved by estimating the distance between stations on the PlanetLab network. In this scope of activity, the application was created to measure the delay prediction using Vivaldi algorithm with adaptive time step which is on principle of operation a client-server where the client performs the steps of the algorithm, Vivaldi and the server only listens, collects the resulting data Vivaldi algorithm and stores them neatly file. Furthermore, the application was developed for direct measurement of the delay, which also functions as a client-server. These applications have been transferred to the selected nodes from the PlanetLab experimental network. Subsequently, these nodes were running, to carry out the necessary measurements. The resulting values were work into tables of using Microsoft Excel. These values were then compared with direct measurements and competitive positioning by the King. Vivaldi localization methods with adaptive time step and the King, were compared based on calculated estimates of both real estate errors and measurement using distribution function of the relative errors of both methods. All this information was evaluated to compare accuracy of both the localization methods and direct measurements.
|
guest ::
